Molecular Microbiology (2002) 43(3), 641–652

Translational readthrough of the PDE2 stop codon modulates cAMP levels in Saccharomyces cerevisiae Olivier Namy,1† Guillemette Duchateau-Nguyen2‡ and Jean-Pierre Rousset1* 1 Laboratoire de génétique moléculaire de la traduction, and 2Laboratoire de bioinformatique des génomes, Institut de Génétique et Microbiologie, Université Paris-Sud, 91405 Orsay Cedex, France. Summary The efficiency of translation termination in yeast can vary several 100-fold, depending on the context around the stop codon. We performed a computer analysis designed to identify yeast open reading frames (ORFs) containing a readthrough motif surrounding the termination codon. Eight ORFs were found to display inefficient stop codon recognition, one of which, PDE2, encodes the high-affinity cAMP phosphodiesterase. We demonstrate that Pde2p stability is very impaired by the readthrough-dependent extension of the protein. A 20-fold increase in readthrough of PDE2 was observed in a [PSI+] as compared with a [psi– ] strain. Consistent with this observation, an important increase in cAMP concentration was observed in suppressor backgrounds. These results provide a molecular explanation for at least some of the secondary phenotypes associated with suppressor backgrounds. Introduction Over the last decade, data have demonstrated that the genetic code may be more flexible than previously supposed. For example, the ribosome can change the decoding frame during elongation, either forward or backward, read through a stop codon, or even skip a part of the message by hopping (Huang et al., 1988; Herr et al., 2000). These alternative readings of the genetic code have been called ‘recoding’ by Atkins and Gesteland (Gesteland et al., 1992). In many cases, these mechanisms result in the bypass of a stop codon and the synthesis of a longer polypeptide than that produced by conventional decoding. Accepted 22 October 2001. *For correspondence. E-mail rousset@ igmors.u-psud.fr; Tel. (+33) 1 69 15 50 51; Fax (+33) 1 69 15 46 29. Present addresses: †Division of Virology, Department of Pathology, University of Cambridge, Tennis Court Road, Cambridge CB2 1QP, UK. ‡Hoffmann-La Roche Ltd, Building 69/338, CH-4070 Basel, Switzerland.

© 2002 Blackwell Science Ltd

Many viral families use programmed –1 (minus one)-ribosomal frameshifting to synthesize the Gag-Pol polyprotein (Brierley et al., 1987; Jacks et al., 1988a; b; Hyun et al., 1993). In the yeast Saccharomyces cerevisiae, the retrotransposons Ty1 and Ty3 are the best characterized ribosomal +1 frameshift sites (Belcourt and Farabaugh, 1990; Farabaugh, 1993). To date, few cellular genes have been shown to regulate gene expression via a frameshifting event. The first example was described in Escherichia coli, in which autoregulation of a programmed +1-ribosomal frameshift in the prfB gene is required for the synthesis of release factor 2 (Craigen et al., 1985; Craigen and Caskey, 1986). Also in E. coli, a –1-ribosomal frameshift in the dnaX gene generates the DNA polymerase g-subunit (Flower and McHenry, 1990; Tsuchihashi and Kornberg, 1990). In yeast, two cellular genes require a +1 frameshift for expression: EST3, encoding a subunit of the telomerase enzyme (Morris and Lundblad, 1997), and ABP140, encoding a new actin filament-binding protein (Asakura et al., 1998). In multicellular organisms, programmed +1-ribosomal frameshifting has been demonstrated in the gene encoding the ornithine decarboxylase (ODC) antizyme; (Matsufuji et al., 1995; 1996; Zhou et al., 1999; Ivanov et al., 2000). Recoding signals can also reprogramme the message of stop codons, either for a canonical amino acid by a classical suppression-like mechanism or, in some particular cases, for the modified amino acid selenocysteine (Martin et al., 1996). Moloney Murine Leukemia Virus uses stop codon readthrough rather than a –1 frameshift to express its Gag-Pol precursor protein. In Sindbis virus, the weak UGAC context permits a 10% readthrough efficiency (Li and Rice, 1993). The context CAA UAG CAA UUA from Tobacco Mosaic Virus (TMV) promotes high readthrough levels in eukaryotic cells, ranging from 2% in mouse cells (Cassan and Rousset, 2001), 5% in plant cells (Skuzeski et al., 1991) to 25% in yeast (Bonetti et al., 1995; Stahl et al., 1995; Namy et al., 2001). The best studied examples of programmed redefinition of a stop codon in cellular genes involve the synthesis of selenoproteins. Selenocysteine incorporation occurs by readthrough of a UGA stop codon, and is stimulated by the presence of a particular mRNA structure (SECIS element) (Martin et al., 1996). There are a few reports of readthrough in other cellular eukaryotic genes, such as the headcase (hdc) gene of

642 O. Namy, G. Duchateau-Nguyen and J.-P. Rousset Table 1. Description of the readthrough motifs. Readthrough motif

ORF namea

ORF1 length (nt)

CAA TAG CAA CAA TAA CAA

YOR360c (PDE2) YDR411c

1581 1026

63 60

AAA TAG CAA AAA TAA CAA

YCL010c YIL036w (CST6)

780 1764

108 75

ATA TAA CAA ATA TAA CAA

YBR027c YCR101c

333 549

90 114

GAA TAG CAA

YLR248w (RCK2)

1833

117

TTT TGA CAA

YBL070c

321

72

AAG stop CAA ATC stop CAA TTC stop CAA

No ORF with 3¢ extension longer than 60 nt contained this motif No ORF with 3¢ extension longer than 60 nt contained this motif No ORF with 3¢ extension longer than 60 nt contained this motif

3¢ extension length (nt)

a. Gene name is indicated in italic when it is known.

Drosophila melanogaster (Robinson and Cooley, 1997; Steneberg et al., 1998; Bischoff et al., 2000). The hdc gene is implicated in different developmental processes, and its sequence reveals a 3241-bp ORF, interrupted by a single in-frame stop codon. The downstream ORF is essential for the production of a functional product. Recently, it has been described that the readthrough mechanism probably involves a secondary structure downstream the stop codon (Steneberg and Samakovlis, 2001). Thus, as in the case of frameshifting, readthrough events have also been observed mainly in small genetic elements such as transposons and viruses. The goal of this study was to identify genes in S. cerevisiae that are partially controlled by a readthrough mechanism. One major parameter known to affect the efficiency of translation termination is the local sequence context surrounding the termination codon. Statistical analyses of genes in many species have shown that nucleotide distribution around the stop codon is not random (Brown et al., 1993). The upstream sequence context affects translation termination efficiency in E. coli (Mottagui Tabar and Isaksson, 1997) and S. cerevisiae (Mottagui-Tabar et al., 1998). Sequences immediately 3¢ of the stop codon are also involved in the efficiency of translation termination (McCaughan et al., 1995; Poole et al., 1995; Tate et al., 1995). Upstream and downstream sequences surrounding the stop codon act synergistically on the efficiency of translation termination in yeast (Bonetti et al., 1995). In these studies, numerous readthrough motifs were tested, and several have been shown to decrease translation termination efficiency of reporter genes. A computer analysis allowed us to identify eight genes bearing one of such previously described readthrough motifs. We demonstrate that a high readthrough level occurs at these sites. In the case of the PDE2 gene, whose product is a key regulator of the cAMP control

cascade, we observed that the carboxy terminal extension, generated by the readthrough event, significantly alters protein stability, resulting in increased cAMP intracellular concentration and a modification of the stress response. Thus, regulation of termination readthrough is one means by which gene activity can be controlled.

Results Presence of readthrough motifs in the genome of S. cerevisiae Eight motifs promoting very efficient stop codon suppression in a sup45 mutant strain were selected for a computer search in the S. cerevisiae genome. Among these, seven were described previously by Stansfield, and the eight by Fearon (Fearon et al., 1994; Stansfield et al., 1995). All are related to the readthrough sequence CAA UAG CAA, initially found in TMV. Only ORFs containing a potential 3¢ extension, with a minimal length of 60 nucleotides before the next stop codon, were retained for further analysis. As indicated in Table 1, eight ORFs bearing one out of five of the eight possible readthrough motifs were identified.

Biological functions of ORFs with a readthrough motif PDE2 encodes the high affinity cAMP phosphodiesterase, which catalyses the hydrolysis of cAMP to AMP (Wilson and Tatchell, 1988). This protein is implicated in the Ras/cAMP pathway, and indirectly modulates the ubiquitin ligase activities of the anaphase-promoting complex (APC) and the Skp1p-Cdc53p-F box complex (SCF) (Irniger et al., 2000). It has been suggested that Pde2p might protect the cAMP pathway from interference as a result of the extracellular environment (Wilson et al., 1993). PDE2 is a non-essential gene, but a null mutant is sensitive to oxidative and heat stress. © 2002 Blackwell Science Ltd, Molecular Microbiology, 43, 641–652

Translational readthrough and cAMP lacZ

have no known motifs, other than the protein encoded by the YCR101c ORF (182 amino acids) containing a 169amino-acid region with 52% identity to Vth1p (a receptor that binds to the integral membrane implicated in vesicular transport) upstream of the stop codon, and 62% similarity to Yip1p (a Golgi integral membrane protein that binds to the transport GTPase Ypt1p and Ypt3p proteins) downstream of the stop. YCR101c is identified at the YPDTM database as a fragmented coding region of a pseudogene.

luc MscI GCTGGCCAA A

643

pAC99 parent

TAA (NNN) 2-6 TAG (NNN)2-6 Readthrough motifs TGA

Fig. 1. Schematic representation of the reporter system. The pAC dual reporter consists of a fusion lacZ-luc gene in which coding sequences are separated by a MscI restriction site. Cloning of oligonucleotides carrying a readthrough motif is illustrated.

Quantitation of stop codon readthrough frequency

RCK2 (CLK1) encodes a calcium/calmodulindependent serine/threonine protein kinase. The Cterminal segment of RCK2 is a negative regulatory domain (Melcher and Thorner, 1996). Rck2p can specifically phosphorylate the translation elongation factor 2 (EF-2) in vitro (Melcher and Thorner, 1996). Hyperphosphorylation of EF-2 blocks the ability to promote a shift of the aminoacyl-peptidyl-tRNA from the A site to P site, and thus totally inhibits translation (Donovan and Bodley, 1991; Brigotti et al., 1992). Rck2p is also a non-essential gene but shares 46% similarity with another calcium/ calmodulin-dependent serine/threonine protein kinase, Rck1p. CST6 was recently characterized (Ouspenski et al., 1999). Cst6p is involved in chromosome recombination and stability. It contains a potential fos/jun DNA-binding domain. Cst6p contains a region of 300 amino acids displaying 40% similarity with cAMP response element binding proteins (CREBP). For these three characterized genes, the presence of the stop codon was verified by sequencing the amplified genomic fragment. The other five ORFs correspond to new genes that

To determine whether these motifs are actual readthrough sites, the regions surrounding the stop codons were cloned at the lacZ-luc junction of a dual gene (Fig. 1) that allows to specifically report for readthrough efficiency, and shows no interference with other levels of control, such as translation initiation or nonsense-mediated mRNA decay (Bidou et al., 2000). We quantitated readthrough frequency as the ratio of luciferase and b-galactosidase activities, in comparison with an in-frame control (Bidou et al., 2000). This was performed in the Y349 wild-type strain, fully active for translation termination and bearing no suppressor tRNA. Results, shown in Table 2, indicate that the eight motifs are sites of readthrough frequency 7 to 31-fold higher than ‘anonymous’ stop codons. In two cases (PDE6wc and CL010.6wc), a region comprising only six nucleotides on either side of the stop codon was tested in the readthrough assay. In both cases, the minimal motif elicited the same readthrough frequency as the initial construct (data not shown). This demonstrates that high-level readthrough can be determined by the immediate neighbourhood nucleotide context of the stop codon.

Table 2. Readthrough quantitation. Sequence

Readthrough frequency a

Control vectors pAC74b pAC-PDEQc

AGC TGA TCA CAA CAG CAA